They’re using ChatGPT. I know.
I know because I can read. Because I’ve been reading the flattened, clunky prose of undergraduates for long enough to recognize a sudden shift when it arrives. Because I, too, have played with the machine, watched it swing between the painfully generic and the eerily precise, the lazy paraphrase and the uncanny conceptual leap. Because there’s a cadence, a soft echo, in the way they write now. Because there are patterns, ticks, favourite turns of phrase, and I have come to know them the way one knows the signature of a plagiarist who thinks nobody’s watching.
So no, I’m not offended. I’m entertained.
For the first time in years, I am not reading with clenched teeth and shallow breath, enduring spelling that defies phonetics, syntax that loops and drowns, claims that dangle like power lines in a storm. Instead, I get sentences. Clauses. Arguments. A flow. An idea, or at least the shape of one. Some are beautiful in a way no teenager should write: clean, controlled, suspiciously lucid. Others are bizarre, like hallucinations put through peer review. A paper on Hamlet as an early practitioner of narrative therapy. A comparison between King Lear and Silicon Valley’s leadership crisis. Shakespeare as the first influencer. Shakespeare as the last. Shakespeare as content.
It’s not that I believe these ideas. I don’t. But I can read them. I can talk about them. They leave traces. They make me think. And that’s more than I can say for the hundred-and-fifty-third limp summary of “theme” in Othello.
What they don’t seem to realize is that we can tell. Of course we can tell. ChatGPT writes like ChatGPT: reliably, recognizably, with a kind of automated composure that no amount of panicked paraphrasing can mask. Its paragraphs rise and fall like factory-made bread. You learn to spot the doughy transitions, the engineered coherence, the syntactic scaffolding that somehow always manages to sound both advanced and generic.
And yet, I let it slide. Or rather: I engage with it. I respond to it. I don’t fail them. Sometimes I even give them good grades.
Why?
Because we need to be honest about what’s really bothering us. It’s not the “cheating.” It’s the loss of control. The sense that something foundational has shifted, not just in what we’re assessing, but in what assessment is.
When students used Wikipedia, we told them to use the library. When they Googled, we told them to think. When they started using Grammarly, we told them to learn grammar. But now that they can summon a whole essay in sixty seconds, we no longer know what to tell them to do instead. Write it “themselves”? What does that mean, exactly? Does anyone write by themselves? Is a student who writes with ChatGPT more fraudulent than the student who gets tutored by their lawyer aunt, or edited by a friend, or trained in the kind of elite schools where essay-writing is practically professionalized?
There’s a lie embedded in the outrage. The lie that university work was ever truly a measure of individual, unassisted, unaugmented intelligence. The lie that what we grade is their thinking, rather than the surface traces of a system that already filters who can articulate what, and how.
If I taught math, and a student used a calculator to arrive at the right answer, would I flunk them for not showing their long division? Or would I put a checkmark and move on? The point was never the method; it was the outcome. Except—was it?
Here, in the humanities, the work is the method. Or so we say. But what happens when the calculator starts generating not just answers, but interpretations? When it starts producing language, rhythm, narrative? When it fumbles and hallucinates and still, somehow, ends up more interesting than most of the real students we’ve had in the last decade?
That’s the hard truth. Sometimes the machine is just better. Not smarter. But faster, clearer, more unpredictable, less afraid. It can fake confidence better than our students can fake knowledge. It can invent. It can lie with panache. It can mimic curiosity, even when our students have forgotten what that felt like.
So now I read these essays, and I ask myself: am I reading them, or the machine? Is this plagiarism, or collaboration? Is it a betrayal of education, or a redefinition of it? I don’t know. But I’m not going to pretend I’m not intrigued.
Maybe the assignment was the problem. Maybe I need to ask better questions—questions the bot can’t answer cleanly, or that push students to make the machine wobble, glitch, fail in revealing ways. Maybe the task now is to teach them to read the machine, not just use it. To annotate its hallucinations. To intervene in its language. To make it strange again.
Because no, I don’t want an AI-generated paper on how Romeo and Juliet is a timeless tale of love and fate. But I might want a student to take that very phrase—“timeless tale of love and fate”—and tear it apart, expose its hollowness, mock its ubiquity, ask why the machine reached for it and what that says about us. That’s a real paper. That’s literature.
So I keep assigning essays. I keep receiving the bot’s faint carbon copies. I keep reading. I keep grading, but I also keep laughing, thinking, watching. The game has changed. Maybe that’s good. Because for once, I’m not bored.